Assessment of earthquake-induced landslide hazard zoning using the physics-environmental coupled Model

In order to prevent and mitigate disasters, it is crucial to immediately and properly assess the spatial distribution of landslide hazards in the earthquake-affected area. Currently, there are primarily two categories of assessment techniques: the physical mechanism-based method (PMBM), which considers the landslide dynamics and has the advantages of effectiveness and proactivity; the environmental factor-based method (EFBM), which integrates the environmental conditions and has high accuracy. In order to obtain the spatial distribution of landslide hazards in the affected area with near real-time and high accuracy, this study proposed to combine the PMBM based on Newmark method with EFBM to form Newmark-Information value model (N-IV), Newmark-Logic regression model (N-LR) and Newmark- Support Vector Machine model (N- SVM) for seismic landslide hazard assessment on the Ludian Mw 6.2 earthquake in Yunnan. The predicted spatial hazard distribution was compared with the actual cataloged landslide inventory, and frequency ratio (FR), and area under the curve (AUC) metrics were used to verify the model’s plausibility, performance, and accuracy. According to the findings, the model’s accuracy is ranked as follows: N-SVM>N-LR>N-IV>Newmark. With an AUC value of 0.937, the linked N-SVM was discovered to have the best performance. The research results indicate that the physics-environmental coupled model (PECM) exhibits accuracy gains of 46.406% (N-SVM), 30.625% (N-LR), and 22.816% (N-IV) when compared to the conventional Newmark technique. It shows varied degrees of improvement from 2.577% to 12.446% when compared to the single EFBM. The study also uses the Ms 6.8 Luding earthquake to evaluate the model, showcasing its trustworthy in forecasting power and steady generalization. Since the suggested PECM in this study can adapt to complicated earthquake-induced landslides situations, it aims to serve as a reference for future research in a similar field, as well as to help with emergency planning and response in earthquake-prone regions with landslides.